| Summary: | 碩士 === 國立中央大學 === 大氣科學學系 === 107 === Model for Prediction Across Scales-Atmosphere (MPAS-A) is a new global non-hydrostatic atmospheric model, which uses unstructured variable resolution meshes with smoothly varying mesh transitions, well suited for a higher-resolution mesoscale atmosphere simulation. For typhoon prediction, observations over ocean are important for a numerical model to better represent the real atmosphere. Global Positioning System (GPS) radio occultation (RO) data have characteristics of global coverage, high vertical resolution, and high accuracy, so they can provide useful information for the data-sparse region.
In this study, GPSRO data are assimilated with GSI, and the analysis is provided as the initial condition for the MPAS-A model using a new technique. The impact of RO observations to the track forecast of Typhoon Nepartak (2016) is investigated. The study consists of two parts: In the first part, the Central Weather Bureau Grid-point Statistical Interpolation (GFS/GSI) hybrid system is used for the assimilation, and the MPAS model forecasts are conducted with 60-15 and 60-15-3 km resolutions. The typhoon bogus is employed during the assimilation process. The experiments with RO data show better track forecast in both high- and low-resolution configurations, and the analysis fields are comparable with the ECMWF analysis. However, these results are obtained using one-way connection from GSI to MPAS; i.e., the data assimilation cycles are not conducted using the MPAS model itself. We tried to build the two-way assimilation system, but we encountered some serious technical problems.
To solve the two-way issue, in the second part, we successfully set up a brand new MPAS-GSI framework developed by NCAR. The impact of the GPSRO data to the typhoon forecasts are thus studied in both 3DVar and Hybrid methods using the MPAS/GSI system. The five-day forecast results show that the typhoon still makes landfall at Taiwan without the bogus process and that the cases with RO data exhibit better track forecasts. However, the track errors in these experiments are not as accurate as those in the first part. As this new two-way system is still at the testing stage, there is still a lot of room for improvement, and more developments can be expected in the future.
|
|---|
